Intel: Unleash the Aldritch Horror 

Love their faces

They show the slide with the Coremark, Sysmark and UL Overall benchmarks for less than 3 SECONDS. Highly suspect that they score a 32% and 36% uplift in pugetbench (which intel bribed to rig their test) but only 19% in CoreMark and 15% in SysMark and the UL OVERALL Benchmarks. THAT is definitely the more important slide. This is not looking good for PugetBench and Intel, BUT I am ALSO quite disappointed in Level1 and Wendell for going along with this.

@@excitedbox5705 lmfao LMFAO if you knew anything you would know to wait for these guys to get their own review up before paying attention to the slides. Yes they give a good idea of what's coming but how have slides gone for the last ten years? They use one specific computation and then judge their whole processor off of it and it's not realistic. Just like it's not realistic for you to think most/any of this. You don't have many friends do you?

@@LoveToMix dude they get so excited or so dejected and I enjoy these guys a lot

LOL to many lies, AMD got the 3rd patch 2 weeks ago, 11 performs exactly the same as 10 now,

Itanium was a nightmare way of trying to create a closed system of 64-bit processing. It had issues that killed it even without it lacking older x86 support.

@@somehow_not_helpfulATcrap There was plans for high end IA64 desktops. For normal desktops Intel planned to continue using IA32. Intel didn't see a future for the laptop market at the time as they were already developing, and released a couple years later, the IA32 based Intel Atom for netbooks.

@@SaturnusDK 'Intel didn't see a future in the Laptop market' just like how they didn't see a future in the iPhone? Whoever was leading Intel back then really had amazing foresight /s

@@carl8790 They saw a future for netbooks, or what we today would call ultrabooks, so in a sense they were on the right track. They just severely overestimated how powerful their IA32 Atom processors would end up being compared to Core architecture based chips they developed as a direct response to AMD64.

@@SaturnusDK oh I didn't even know that. That's even more hilarious. Thanks for the history lesson.

Its not Beethoven but ok (see description) :D

@@DoctorrMetal This is most certainly Beethoven's Egmont overture. I love the piece to bits. Along with the Coriolanus overture.

@@DoctorrMetal it's beethoven's egmont overture (which is in the public domain), performed by Kevin Macleod. :)

quoting sifi movies brings all ladies to the yard

Because an 8k monitor sucks 400w constantly.

I really dont want to have to spend 60$ a month on AC just to keep my gaming room below 35C in winter. I used to run a 1090T and a HD6950 "was round 600watts total" with an old plasma 45 "IIRC" inch plasma tv for a monitor "another 3-400 watts." I did not have to turn on a heater all winter and winter is cold here. So an Intel 12900k "upto 250watts stock" With a 3090 yeah it would be about the same.

3090 uses over 400W of power. I dont care if my CPU uses 150W, its meaningless when my GPU is significantly worse

@@dralord1307 1kw/h and you would be bankrupt where I live

@Matthew Shields some testing here ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-8hIbG5cTWUs.html

Its already in windows and the patch worked so yes AMD is up to speed. But that matters not to performance enthusiasts concerning. Intel threw everyone a curve ball, Intel has 2 stock configurations. Depending on your cooling . 125 watt PL1/241 watt PL2 they are both listed as stock setting , I want Full Turbo 241 watts , GPUs will be over 450 watts soon and are over 300 watts now, So skip the song and dance , Stock is what intel says it is not AMD fanbios. This is going to be fun . The left will cry and cry time and time again.

Someone tested performance after the patch and saw barely any change, within margin of error. Intel told reviewers ahead of time they used the pre-patch AMD numbers because AMD only patched it a few days before the event, which didnt give Intel enough time to do benchmarks again. Also its kind of AMD's fault they took over 3 weeks to fix an issue for windows 11, while Intel didnt have issues.

I don't think it's AMDs fault windows 11 is a rushed public beta test that is being pushed out way too early. Did microsoft not even bother testing with an AMD CPU?

@@__aceofspades since when is a windows issue on AMD to fix?

You will be very happy and the VFM factor is great. Same platform adopted 5 months back. Will run on it till late 2023. But am excited that Intel is back. Once Mid generation these products will have great pricing after the intial flush and hype. I hear that the mobos and ram will put a bigger dent on your budget allocation. So we may have to wait till next year to see reasonable pricing especially in items like ram. Hope we will see availability issues get better next year. Good to see the CPU manufacturers really innovating.

Yeah no regrets, great choice for the next maybe 5+ years?

@@SianaGearz Well it's replacing a 3770k from 2013 that's been running every app and game I wanted it to at 4.3ghz.. I'm aiming for a bit more than 5 years :)

Is this a problem you have with L1 or with intel?

@@adammills4099 Intel's a big offender, but it's an industry wide problem.

Generally we agree but not everyone is aware of the new p core/e core thing and that's worth talking about now so the launch vid isn't so long

As if the embargo nonsense changes the end product. It'll still be as good or crap with or without this bs.

@@adammills4099 Intel specifically. Level 1 made the best of it to explain the tech, rather than just becoming part of the Intel extended PR machine

And 5800x is probably almost the same if you ignore multi core to save even more

Probably thank your crazy government for that….

@@Yooshist Bruh we literally pay less tax than these than USA, also our federal government is fucking useless. If I could power a CPU directly from burning coal I would probably qualify for a subsidy so long as I planted a tree.

Something people hardly consider when comparing AMD lineup to Intel, Intel motherboards are still a lot more expensive than AMD counterparts. Their performance chipsets are expensive, their CPUs are power hungry which requires expensive VRM solutions so expect a $100-150 mark up for any motherboard. Meanwhile you can get amazing B550 motherboards for almost half the price of an amazing Z590 board. Alder Lake does nothing to change that so I expect the same. They also need much more expensive cooling solutions (360 high end AIOs) when a $50-70 dual tower heatsink does the job for an AMD part, even at 16 cores.

@@ufsteropolstero6014 This just isn't the case in Australia. High end z590 boards are priced the same as high end b550 here, although the price ceiling on intel boards is much higher usually. Currently I can go buy an 11700k and a mid-range z590 board for 700 AUD, where the 5900x costs ~750 AUD

There hasnt been much change overall with L1 that us normal ppl could understand. There has been some prefetch changes and stuff "AMD did a huge cache redesign for the 3k series." Where as L3 can end up proving a lot more performance and is a bit easier to the normal ppl to understand "IE MOREEE is better"

@@dralord1307 Exactly. The Zen3D announcement was literally "we slapped some extra L3 on top of existing chips, and got 15-25% improvement in games. You're welcome."

The speed of finding data in cache is generally inversely related to the size of the cache, so L1 cache is going to stay roughly the same size across generations and SKUs to maintain low latency. L2 you're a little less worried about latency so that will vary in size a bit, and L3 is "at least it's not system memory" so make it as large as you have space for.

@@cream_soda one would hope that at some point an improvement of L1 would let you have bigger size same latency or just faster... and even better across all the caches, but my understanding is that this has pretty much hit physical limits very quickly (except for L3 which will keep growing)

I thought about it, but I didn't want to cause any insanity since it could effect our views. ~Editor Autumn

The K models will all run at full boost by default. Basically now MCE is stock and the i9 is practically a 241W cpu, with the i5 being 150W.

Move to France that uses nuclear energy and the price for kwh is literally half.

It will run at like 20W for normal office use. The 240W is only hit when rendering and similar heavy all core tasks.

None of these CPUs will run over 200W in normal workloads, The i9 12900k will only do around 210W with 100% synthetic loads like cinebench, in real world uses like gaming, nowhere close to that.

@@mrfrenzy. Only overclocked with PL2 power limit. Turbo Boost is not infinite.

According to Wikipedia and it's source, Spectre and Meltdown are already resolved.

Been patched completely in hardware. Now AMD is facing its own attacks because they are actually selling CPUs. Previously no security analysts were probing AMD CPUs because in the bulldozer and early Zen days they had like 10% market share

You mean Ryzen 4, because if it can't compete the market will become just Intel once again.

@ I don't care who is on top. I want competition on the market, as that pushes innovation and we as customers get to profit the most. Also, AMD would be no different if they didn't have competition. I'm currently using a 5900X btw.

@ The 3d cache chips coming are supposed to see 15-25% improvement simply from the cache. Im not that worried that AMD's next gen will compete. The bigger question will be pricing, support and additional features I think.

AMD's current gen is competing. Production benchmarks are yet to come and Alder Lake isn't looking like a clear winner there. And it isn't even a clear winner in some games. Alder Lake really isn't the formidable champion it was made out to be. However, unless AMD drops prices, Alder Lake looks like a much better deal on the mid to low end (assuming the mainboard and DD5 prices are reasonable enough).

@@darreno1450 Pretty sure it will take a good while until DDR5 prices are in parity with DDR4 just like when Broadwell and DDR4 launched(the reason why I went with Haswell refresh and DDR3 for my current PC, not to mention they only released 2 CPUs for the desktop). Also, current-gen AMD needs to do more than compete, it needs to dominate, look at how long it took for Intel to get off its ass and innovate, it took as long as it did AMD to reach parity in gaming performance.

How the turns have tabled :V

While Intel will use chiplets in their 2023 CPU lineup (Meteor Lake), Alder Lake uses a monolithic die, so your comment does not make sense!

Makes more sense than p-cores, current Xeons using big cores have bad singlethread perfomance compared to the Core series anyway, (Skylake at 5GHz, Cypress Cove at 5GHz, Willow Cove, etc).

The 12k series will beat the current R5k series. The question is power usage, total system costs, and how the next gen AMD chips turn out.

They show the slide with the Coremark, Sysmark and UL Overall benchmarks for less than 1 SECOND. Highly suspect that they score a 32% and 36% uplift in pugetbench (which Intel has "funded") but only 19% in CoreMark and 15% in SysMark and the UL OVERALL Benchmarks. THAT is definitely the more important slide. This is not looking good for PugetBench and Intel, BUT I am ALSO quite disappointed in Level1 and Wendell for going along with this.

@@dralord1307 Gaming for sure, but that really only matters at 1080P (maybe 1440p). Production is where I have doubts Alder Lake will have a clear lead. The 5950 is a beast and has a lot of head room for tweaks.

While I would've liked to have seen Intel kept in the doghouse for another generation as punishment for their decade of quad cores, I agree that competition is ultimately the better outcome for everyone, so I too am looking forward to seeing what AMD respond with.

@@darreno1450 I fully agree with that. But gonna wait for RL testing before saying which is better. I do think proably Intel will have the lead in games but AMD where you need max thread horsepower

I think he is more excited about the fat check intel slipped them for this. They show the slide with the Coremark, Sysmark and UL Overall benchmarks for less than 1 SECOND. Highly suspect that they score a 32% and 36% uplift in pugetbench (which Intel has "funded") but only 19% in CoreMark and 15% in SysMark and the UL OVERALL Benchmarks. THAT is definitely the more important slide. This is not looking good for PugetBench and Intel, BUT I am ALSO quite disappointed in Level1 and Wendell for going along with this.

@@excitedbox5705 You've got problems. Who cares if intels' data is wrong. We'll find out in a few days anyways after enbargo lifts. Just enjoy the content.

Why not though? 8 e cores are certainly going to boost mutlicore performance vs 2 P cores. These e cores take up 1/4 the die space of a P core while performing better thab 1/4 a P core

You need to do more research then. Efficiency cores aren't just added for power efficiency, but because for the size of 1 P-core you can get 4 E-cores instead (important see below), and we have now found out those cores are similar to 10the gen core performance. So as we've seen in the leaked cinebench benchmarks and now the official Intel slides, Intel is now magically competing with Zen 3 on highly threaded workloads. An insane 50-100% improvement gen over gen in MT. How? E-cores. E-cores are basically an easy way to scale core counts for highly threaded applications. If Intel went with the traditional homogeneous design of only P-cores, Intel would be spending another year trailing AMD in MT performance. 4 E-cores is better than 1 P-Core for MT. I'm sure Intel has the data, but very very very few applications are extremely latency sensitive and need high thread counts. Look at the 5900x and 5950x basically no games have meaningful performance over the 5800x. Games arent actually needing more than 8 cores, most perform extremely well with 6 cores. The apps that do scale extremely well aren't sensitive and will happily use many more cores that are slower individually but overall are faster. So a 8C+8c is better than a 10C and both would have similar die size. For HEDT, productivity and content creation workloads, it actually makes more sense to scale the E-cores than give you more P-cores.

Smart. This is exciting, but I'll let others be the beta testers.

@@benjaminoechsli1941 I was tempted, but yeah I'm thinking if Raptor Lake is just an iterative upgrade, I'll buy a 12700 when they go on discount. I'm also banking on Infel being able to manufacture enough of both CPUs... but I'll take that gamble if it means I can save $200.

It would, in fact, be nowhere as fast, since Zen power scaling is bleh.

It was more a commentary on what Intel can do because they make small space heaters. Of course they can put up impressive numbers but are also consuming a ton more power. I am glad they are innovating though because it is good for everyone. I just wish it was at a lower power consumption level. 🤷‍♂️

​@@pkt1213 These numbers are impressive? Alder Lake will lose in some games and may likely lose in some production workloads. I fail to see what's so impressive here. Where Alder Lake rules is in the mid to low end where 1080P gaming is still popular. And that's mainly because of the lower price point. For 1440p and higher gaming, the difference is a wash.

🤣 Everyone is a critic when you are just trying to poke fun at intel.

@@pkt1213 or because their architecture scales with power better? That still remains an advantage even if you can't see it.

- Vegeta, what does the Wattmeter say about his power usage? - It's over 9000!

over 300 at least. Now add in a 3090 oc'd and we are OVER 9000!

Hell yeah we do, guy fawkes masks are iconic Look at V for vendetta Edit: tbh I don’t know the specifics of the gunpowder plot but I know of it

Lol was going to say naw they either only know of V For Vendetta or Anonymous.

@@Dan-Simms You know, a lot of brits moved to other parts of the world right.

@@dralord1307 the joke was that non Britt's don't know history.

It's amazing that brits know about it, the scheme masterminded by a genius who used the fake name of John Johnson

Only on Linux son... Only in Linux.

not sure its appropriate to use on a company known to do everything in their power to "oppress" the competition, with bribes, "marketing funds" etc

What makes x86 bad?

@@vyor8837 legacy

@@carl8790 what about the legacy? What about having old instructions makes it slower?

Youre wrong though as AMD is moving the the same heterogeneous big.little concept with Zen 5. This is the future and everyone from ARM to Apple to Intel to AMD will be using it. Efficiency cores aren't just added for power efficiency, but because for the size of 1 P-core you can get 4 E-cores instead (important see below), and we have now found out those cores are similar to 10the gen core performance. So as we've seen in the leaked cinebench benchmarks and now the official Intel slides, Intel is now magically competing with Zen 3 on highly threaded workloads. An insane 50-100% improvement gen over gen in MT. How? E-cores. E-cores are basically an easy way to scale core counts for highly threaded applications. If Intel went with the traditional homogeneous design of only P-cores, Intel would be spending another year trailing AMD in MT performance. 4 E-cores is better than 1 P-Core for MT. I'm sure Intel has the data, but very very very few applications are extremely latency sensitive and need high thread counts. Look at the 5900x and 5950x basically no games have meaningful performance over the 5800x. Games arent actually needing more than 8 cores, most perform extremely well with 6 cores. The apps that do scale extremely well aren't sensitive and will happily use many more cores that are slower individually but overall are faster. So a 8C+8c is better than a 10C and both would have similar die size.

Its not really for desktop. Its really for mobile. Intel is just porting it into desktop. In mobile the E cores will help to extend batter life a lot "a place AMD has been kicking Intel hard recently" Intel putting it on desktop can help a lot with their packaging and other tech they need to improve. "IE we are the beta testers for them"

How do you ‘optimize’ for snake oil? Asking for a friend?

Most desktop computers spend most of their time sitting waiting for you. Why not have them able to execute undemanding tasks extremely efficiently? Saves money on power. Saves money on AC. It does require a good implementation of Big/Little in hardware, and an operating system tuned to make good use of both types of cores. Apple has done it, although it does have the advantage of complete control of the hardware and software to make it easier to achieve. But still no reason to think others couldn’t do it too.

@@williamp6800 As LTT already showed in their video on the 12000 series, a optimization issue could occur. I am not saying others can't do it, I just doubt Microsoft can manage to optimize for both this and traditional x86 architecture. As you said your self, what Apple does require complete control of both hardware and software. But I would like to add that it also requires control over the Software (apps) that the Software (IOS) runs. This is where the "pay companies to optimize for something that AMD wont have" comes from. Cause Microsoft don't have that kind of control. (and that's a good thing in my book) But only time will tell I guess.

@@dralord1307 Efficiency cores aren't just added for power efficiency, but because for the size of 1 P-core you can get 4 E-cores instead (important see below), and we have now found out those cores are similar to 10the gen core performance. So as we've seen in the leaked cinebench benchmarks and now the official Intel slides, Intel is now magically competing with Zen 3 on highly threaded workloads. An insane 50-100% improvement gen over gen in MT. How? E-cores. E-cores are basically an easy way to scale core counts for highly threaded applications. If Intel went with the traditional homogeneous design of only P-cores, Intel would be spending another year trailing AMD in MT performance. 4 E-cores is better than 1 P-Core for MT. I'm sure Intel has the data, but very very very few applications are extremely latency sensitive and need high thread counts. Look at the 5900x and 5950x basically no games have meaningful performance over the 5800x. Games arent actually needing more than 8 cores, most perform extremely well with 6 cores. The apps that do scale extremely well aren't sensitive and will happily use many more cores that are slower individually but overall are faster. So a 8C+8c is better than a 10C and both would have similar die size.

In this case, intel's alder lake is much more a response to apple's M1. And it does not look very good for intel, to convince apple to come back to them. Alder Lake is hardly any faster than apple's new M1 iterations.

Its like rooting for Huion over Wacom, who had a monopoly on drawing tablets for eons.

Same here, want them to retake the top, so AMD's next product will be better.

More cores, less price. I doubt they have a big-little concept. And sure they have PCIe5 and DDR5 at one time but currently this doesn't matter.

Looking at intels graphs they have about a 5% performance lead with ddr5 on windows 11 vs amd on windows 10. They are on a 7nm process for the first time which is cool.

@@llothar68 They are going to 5nm no? And Lisa said there are arch changes too.

Ian at Anand asked them. The only games that have issues are ones with old DRM that arent being updated. Out of the top 200 Denuvo DRM games tested only 20 had issues, and Intel was working to fix those

Its Intel 7. which is equivalent to TSMC 7nm. This is a brand new node for Desktops, Intel was previously on 14nm

Thanks for the reply! That's good! Intel's back in the game! Oh right 14nm, my mistake!

In short a CPU like a 5950X tops out 142W stock, the i9-12900K is 241W stock.

@@SolarianStrike Not even close to true.

@@__aceofspades Intel's own marketing slides states that. The K models will run at PL2 by default, that is 241W for the i9 and 150W for the i5. MCE is standard behavior for 12 gen.

Intel Silicon /s

So funny no one laughs 😐

Its Intel 7, equal to TSMC 7. Bad joke though